Key points are not available for this paper at this time.
The optimisation of drug delivery systems is a complex, multidimensional challenge involving the interplay of formulation composition, process parameters, and biological performance. Traditional empirical and statistical approaches are increasingly limited by the high dimensionality, nonlinearity, and multi-objective nature of modern drug delivery problems. In this review, we explore how artificial intelligence (AI) and machine learning (ML) are transforming formulation science by enabling data-driven, adaptive, and efficient optimisation strategies. We provide a conceptual and practical overview of ML-guided optimisation workflows, including surrogate modelling, Bayesian optimisation, active learning, and multi-objective optimisation. Key challenges such as data scarcity, experimental throughput, and model interpretability are discussed. Applications across diverse delivery modalities, including solid oral dosage forms, lipid nanoparticles, biologics, and long-acting injectables, are critically examined, highlighting how ML can accelerate formulation development, reduce experimental burden, and uncover novel design spaces. We conclude by outlining future directions for integrating AI into pharmaceutical R&D, with a focus on the emergence of self-driving laboratories. This review aims to equip drug delivery scientists with the foundational knowledge and practical tools to harness AI and ML in the rational design and optimisation of advanced drug delivery systems.
Ros et al. (Thu,) studied this question.